The Influences of Low-Frequency Vorticity on Tropical Cyclone Formation

Based on Systematic Model SimulationsYi-Huan Hsieh1 , Cheng-Shang Lee1, 2 Chung-Hsiung Sui1

1Department of Atmospheric Sciences,National Taiwan University, Taipei, Taiwan

2Taiwan Typhoon and Flood Research Institute,National Applied Research Laboratories, Taiwan

Oct 4, 2016

Outline1. Introduction2. Environment of Tropical cyclone (TC)

formation in the western N Pacific (WNP) 3. Diagnosis of systematic simulation results 4. Cumulus scheme experiments5. Summary

2

…it is far more natural to assume that genesis is a series of events, arising by chance from quantitative fluctuations of the normal disturbances, with the probability of further evolution gradually increasing as it proceeds.

…the climatological and synoptic conditions do not directly determine the process of genesis, but … affect the probability of its happening.

(Ooyama, 1982)

~100 km ~1000 km ~5000 km

Time

Scale

VHTs,MCVs

- (Ritchie and Holland,1997; Simpson et al., 1997;

Hendricks et al., 2004; Montgomery et al., 2006;

Houze et al., 2010)

Synoptic Environments,Easterly Waves

- (Briegel and Frank, 1996; Ritchie and Holland, 1999; Dunkerton et al., 2009; Montgomery et al., 2009;

Wang et al., 2010a, b; Montgomery et al., 2010; Chang et al., 2010; Lin and Lee, 2011; Wang et al., 2011)

MJO, Kelvin wavesEquatorial Rossby Waves,

- (Roundy and Frank, 2004; Frank and Roundy, 2006; Gall et al., 2010; Gall and Frank, 2010; Ching et al., 2010 )

TC Formation is a stochastic process

Any deterministic natureof TC Formation???

The 32-day ensemble forecast model of ECMWF was able to resolve the formation of most TCs (2009–2010), but some of the weak and short-lived TCs were missed.

- (Elsberry et al., 2010, 2011; Tsai and Elsberry, 2013; Elsberry et al., 2014)

To understand the capability of a numerical model to simulate (forecast) TC formation properly under different environments*1? (*1 Ritchie and Holland, 1999)

5

TCs with Higher low-frequency vorticity, 26 HTCs - HTCsTCs with Lower low-frequency vorticity, 26 LTCs - LTCs

Background 850-hPa vorticity of pre-TC disturbances

LTC HTC

low pass filtered vorticity (×10-5 s-1)

high

pass

filte

red

(×10

-5s-

1 )

(red: 2008–2009, black: 2000–2007)

1.15 × 0−5 s−1

52 TCs in 2008–2009

The 10-day*2 low pass and high pass filters are applied to NCEP_FNL (2000-2009) data to obtained low/high pass filtered winds. - (*2 Wu et al., 2013)

Use filtered winds to compute 850-hPa mean vorticity within 5° radius of the pre-TC disturbance in the WNP at 24-48h before the formation of TC (Vmax ~ 25kt).

Synoptic environments during TC formation (850hPa) - 48 h 0 h (25 kt)

HTCs

LTCs

monsoon-like environment

easterly wave-like system

TCs with higher low-frequency

vorticity

TCs with lower low-frequency

vorticity

Vorticity (~2x 10-5s-1), wind vectors and cloud top temperature

Model setup of systematic numerical simulations

Use WRF V3.2.1 to simulateall 52 TCs in 2008 and 2009

Cloud microphysics: WDM6Cumulus scheme: Kain-FritschPBL Physics : YSU

-Kieu and Zhang, 2008; Chiao and Jenkins, 2010; Wang et al., 2010;Crosbie and Serra, 2014; Li et al., 2014; Xu et al., 2014

Simulations started at 4 distinct initial times (-48h, -72h, -96h, & -120h)T0

Integration times

Initial conditions: NCEP_FNL &EC_YOTC (available only in 2008-2009)

For each TC: 2 (initial conditions) x 4 (initial times) = 8 members416 runs

• Domain 1 : 569 x 340 [36km]• Domain 2 : 706 x 400 [12km]

Target period: -12 hr ~ +12 hr of 1st 25 kt (JTWC best track)

1. Clear circulation center and max. vorticity center at 850 hPa

2. Mean vorticity at 850 hPa:

• > 7.9 * 10-5 s-1 inside 1.5∘

• > 3.8 * 10-5 s-1 inside 3∘, or > 1.5 * 10-5 s-1 inside 5∘

(mean – 1xSD of 52 TCs in EC-YOTC data)

3. Satisfy above criteria for 12 hours or longer

Sugi et al., 2002; Chauvin et al., 2006; Yoshimura et al., 2006; Stowasser et al., 2007; Jourdain et al., 2011; Zhan et al., 2011

Criteria used to defined a model-simulated TC

9

Classify all 416 simulations into 5 groups: no_TC, Simulated_P, large track error_E, large track error_L, and large track error_B

The classification of model simulated TCs

Dashed circle –mean track error ofall simulated TCs(varies with initial time)

-48h ~ 249 km,-72h ~ 301 km,-96h ~ 441 km,

-120h ~ 600 km,

10

Model is more capable of simulating the formation of a HTC, but w/larger location bias. Model is less capable of simulating the formation of a LTC, but w/smaller location bias.

Percentages of five types of simulation results

TC formation Simulated

TC formation Simulated /track error

No TC formation

LTC HTC

- 120h- 96h- 72h- 48h

TC fo

rmat

ion

sim

ulat

ed

Larg

er tr

ack

erro

r

LTC HTC

Expt TC# Cumulus Parameterization

PBL Physics

Micro-physics

Long-wave

radiation

Short-wave

radiation

Initial condi-tions

Initial times

MemberNumber per TC

CTL 52 Kain-Fritsch (new Eta)

YSU WDM6 RRTM Dudhia

EC-YOTC

&NCEP-

FNL

-48, -72, -96, -120 hr

8

CU_EXP 14

Kain-Fritsch (new Eta),Betts-Miller-Janjic,

Grell-Devenyi ensemble,and Grell 3D ensemble

32

• 7 HTCs with Highst low-frequency vorticity (HHTCs): Halong (2008), Kalmaegi (2008), Fung-Wong (2008), Mekkhala (2008), Morakot (2009), Dujuan (2009), and Ketasna (2009)

• 7 LTCs with Lowst low-frequency vorticity (LLTCs): Nakri (2008), Nuri (2008), TS14W (2008), Maysak (2008), Haishen (2008), Noul (2008), and Lupit (2009)

11

Results sensitive to the physics schemes used ?• 14 extreme cases are selected to perform the sensitivity test.

• The cumulus scheme appears to be the most important one,4 cumulus schemes are used in the test (CU_EXP).

(376 more runs, totally 792 runs)

12

Cumulus experiment

The relative proportions of five simulation results for HHTCsand LLTCs are similar to those for HTCs and LTCs

TC formation Simulated /track error

No TC formation

TC formation Simulated LLTC HHTC

- 48h - 72h - 96h - 120h

LLTC HHTC

E

L

B

13

Grell 3D ensembleGrell-Devenyi ensembleBetts-Miller-JanjicKain-Fritch

Most simulations (30/32) have “TC formation”, but some of them have large track errors.

KF

BMJ

GD

G 3D

Simulated results of a HHTC (Dujuan, 2009) at T0

850 hPa vorticity

Obs.

Hourly rain rates (mm h-1) at 0625 UTC 03 Sep 2009(TMI/PR from NRL’s website)

Cloud top temperature (K, shaded, gridsat data) &

850-hPa winds and vort at 0600 UTC 03 Sep 2009 (> 5 × 10-5 s-1, red contours at 5 × 10-5 s-1 intervals)

Shadings - simulated composite reflectivity (the maximum reflectivity at grid column)

Observed vs. simulated convective features (Dujuan, 2009)

• Convection patternssimilar to observation,but different in strength

• All have “TC formation”,but with different track errors

• Whether TC will form or not seems to be not too sensitive to cumulus schemes for HHTC

HHTC (T0-12)

-48 hr / EC-YOTC

15

Betts-Miller-Janjic

KF

BMJ

GD

G 3D

Simulations using Betts-Miller-Janjiccumulus scheme generally have bettercyclonic circulation and higher vorticity

Grell 3D ensembleGrell-Devenyi ensembleBetts-Miller-JanjicKain-Fritch

850 hPa vorticity

Simulated results of a LLTC (Nuri, 2008) at T0

-48 hr / EC-YOTC

“TC formation” occurs only in simulations using BMJscheme

Convection pattern is sensitive to the cumulus schemes used for LLTCs

Hourly rain rates (mm h-1) at 0703 UTC 16 Aug 2008(TMI/PR from NRL’s website)

Cloud top temperature (K, shaded, gridsat data) &

850-hPa winds and vort at 0600 UTC 16 Aug 2008(> 5 × 10-5 s-1, red contours at 5 × 10-5 s-1 intervals)

Shadings - simulated composite reflectivity (the maximum reflectivity at grid column)

Observed vs. simulated convective features (Nuri, 2008)LLTC (T0-12)

Higher low-frequency vorticity

• monsoon-like• easier for TC formation

(higher percentage)• larger track error• Simulation results not too

sensitive to the cu schemes

Lower low-frequency vorticity

• easterly wave-like• harder for TC formation

(lower percentage)• smaller track error• Simulation results sensitive to

the cumulus schemes

Summary

The convection process (cumulus scheme) is not the key factor for TC formation in an environment with large low-frequency vorticity, but very important if the environmentallow-frequency vorticity is small.

Conclusions• Under an environment (monsoon-like) with favorable

condition, especially large vorticity, TC formation can be expected -- deterministic nature of TC formation.

• But the timing and location of TC formation are affected by convective process -- stochastic nature.

• Under a less favorable environment (easterly wave-like), convections play key role to TC formation-- TC formation is more like a stochastic process.

(?)

Need more studies to fully address this issue.

20

TIGGE forecasts (ECMWF, NCEP, AMMC)The THORPEX Interactive Grand Global Ensemble (TIGGE)

The relative proportions of five simulation results in TIGGE forecasts are similar to those for HTCs and LTCs

TC formation Simulated /track error

No TC formation

TC formation Simulated LTC HTC

- 48h - 72h - 96h - 120h

LTC HTC

EL

B

low pass filtered vorticity (×10-5 s-1)

high

pass

filte

red

(×10

-5s-

1 )(red: non-formation, black: formation)